Spending 2020 under the shadow of a pandemic has affected what we need and expect from technology. For many, COVID-19 accelerated the rate of digital transformation: as employees worked from home, companies needed AI systems that facilitated remote work and the computing power to support them.
The question is, how should companies focus their resources in 2021 to prepare for this changed reality and the new technologies on the horizon? Here are three trends that I predict will see massive attention in 2021 and beyond.
1. AI must become practical
Progress in AI has already reached a point where it can add significant value to practically any business. COVID-19 triggered a massive sense of urgency around digital transformations with the need for remote solutions. According to a report by Boston Consulting Group, more than 80% of companies plan to accelerate their digital transformation, but only 30% of digital transformations have met or exceeded their target value.
Many AI projects are small scale — less than a quarter of companies in McKinsey’s 2020 State of AI reported significant bottom-line impact. This is especially true in industries that have a physical-digital element. For example: There is a great need for remotely operated, autonomous manufacturing facilities, refineries, or even, in the days of COVID-19, office buildings. While the underlying technology is there, achieving scalability remains a concern and digital leaders will have to overcome that barrier in 2021. Scalability barriers include a lack of disciplined approach, enterprise-wide mindset, credible partners, data liquidity, and change management.
Part of the solution here is to create solutions that will be operated by someone who is not necessarily a data scientist, so more people who are domain experts can manage the programs they need. If Tesla invented an autonomous car that only data scientists can drive, what’s the point?
Technology needs to empower the end user so they can interact with and manipulate models without having to trudge through the finer points of datasets or code — in other words, the AI will do the heavy lifting on the back end, but a user-friendly explanation and UI empowers the end user. For instance, a facilities management executive can manage their global portfolio of buildings from a tablet sitting at a Starbucks. They can have full visibility into operations, occupant experience, and spend, with the ability to intervene in what otherwise would be an autonomous operation.
2. Solutions become more autonomous with deep learning
Deep learning pioneer Dr. Geoffrey Hinton recently told MIT Technology Review that deep learning will be able to do “everything” — i.e. replicate all human intelligence. Deep neural networks have demonstrated extraordinary capabilities to approximate the most relevant subset of mathematical functions and promise to overcome reasoning challenges.
However, I believe there is a step to full autonomy that we must first conquer: what Dr. Manuela Veloso at Carnegie Mellon calls symbiotic autonomy. With symbiotic autonomy, feedback and correction mechanisms are incorporated into the AI such that humans and machines pass information to each other fluidly.
For example, instead of hard feedback (like thumbs up and thumbs down powering your Netflix queue), symbiotic autonomy could look like a discussion with your phone’s virtual assistant to determine the best route to a destination. Interactions with these forms of AI would be more natural and conversational, with the program able to explain why it recommended or performed certain actions.
With deep learning, neural networks approximate complex mathematical functions with simpler ones, and the ability to consider a growing number of factors and make smarter decisions with fewer computing resources gives them the ability to become autonomous. I anticipate heavy investment in research of these abilities of deep neural networks across the board, from startups to top tech companies to universities.
This step toward fully autonomous solutions will be a critical step towards implementing AI at scale. Imagine an enterprise performance management system that can give you a single pane of visibility and control across a global enterprise that is operating multiple facilities, workers, and supply chains autonomously. It runs and learns on its own but you can intervene and teach when it makes a mistake.
3. Promise of curing future pandemics will accelerate research in quantum computing
Quantum computers have the computational power to handle complex algorithms due to their abilities to process solutions in parallel, rather than sequentially. Let’s think of how this could affect development and delivery of vaccines.
First, during drug discovery, researchers must simulate a new molecule. This is tremendously challenging to do with today’s high-performance computers, but is a problem that lends itself to something at which quantum computers will eventually excel. The quantum computer could eventually be mapped to the “quantum system” that is the molecule, and simulate binding energies and chemical transition strengths before anyone ever even had to make a drug.
However, AI and quantum computing have even more to offer beyond creating the vaccine. The logistics of manufacturing and delivering the vaccine are massive computational challenges — which of course makes them ripe for a solution that combines quantum computing and AI.
Quantum machine learning is an extremely new field with so much promise, but breakthroughs are needed to make it catch investors’ attention. Tech visionaries can already start to see how it’s going to impact our future, especially with respect to understanding nanoparticles, creating new materials through molecular and atomic maps, and glimpsing the deeper makeup of the human body.
The area of growth I am most excited about is the intersection of research in these systems, which I believe will start to combine and produce results more than the sum of their parts. While there have been some connections of AI and quantum computing, or 5G and AI, all of these technologies working together can produce exponential results.
I’m particularly excited to see how AI, quantum, and other tech will influence biotechnology as that might be the secret to superhuman capabilities — and what could be more exciting than that?
